Jose A. Bernat Bacete/Getty Images

The latest iteration of Google’s video-generating AI model, Veo 3, continues to evolve at a rapid clip.

As part of its latest upgrade, the model now lets users generate eight-second video clips, including AI-generated audio, from a single still image. According to Google Cloud documentation updated Monday, the feature is now available as a “preview offering.” Josh Woodward, head of Google Labs and the Gemini App, originally wrote in an X post last week that the company was working on image-to-video capabilities for Veo 3.

What to use Veo 3 for

An influencer, for example, could upload a single headshot of herself and prompt the model to generate a short clip of her walking down a runway wearing a product from a brand she’s partnered with. Veo 3 would automatically include ambient noise, like the murmurings of the crowd and her footfalls on the floor; the user could also request that her AI-generated likeness speak a few lines, like in this example.

Also: This new AI video editor is an all-in-one production service for filmmakers – how to try it

Brands could also use the new feature by feeding the model an image of a product and asking for a clip that displays it from a range of different perspectives. Amazon has developed an AI tool for advertisers with similar capabilities, while Meta has vowed to go further, stating its plans to automate the entirety of the ad-production process.

Veo 3’s new image-to-video capability could help creative professionals in various industries save time and resources that would otherwise be spent organizing on-site video shoots. It can also provide more creative materials for use across social media and other channels.

Google revealed Veo 3 in May at its annual I/O developer conference. The model quickly attracted attention from AI researchers and creative professionals for its ability to seamlessly integrate AI-generated video and audio, a highly technically complex feat that promises to open new doors for AI-assisted filmmaking. It also excels at simulating real-world physics and isn’t hindered by the many technical glitches that plagued earlier AI-generated video tools.

Also: How AI companies are secretly collecting training data from the web (and why it matters)

There’s no sign that Google’s investment in Veo 3 will slow down anytime soon. Last week, Google DeepMind CEO Demis Hassabis appeared to hint in a X post that the model could soon be used to generate virtual worlds for video games. The timing of that prediction that is interesting, given Microsoft laid off 9,000 people from its gaming division earlier this week.

How to try it

Originally only available through Gemini Ultra and Flow, Veo 3 was generally released as a public preview last month — all Google Cloud customers and partners can access it in the Vertex AI Media Studio. The model is now available across 159 countries.

Controversy and potential risks

Veo 3 has sparked concerns over AI’s potential to supercharge the spread of online misinformation and manipulate users on social media. There are also questions around the sourcing of its training data, which Hassabis has said could include YouTube videos.

Also: This open-source bot blocker shields your site from pesky AI scrapers – here’s how

Since AI companies scraped much of the text, image, audio, and video content they use to train their models from the open internet, creators from across the publishing, art, and film industries have raised copyright issues with these generators. If you’re looking for a more airtight AI-generated video tool, consider checking out Moonvalley’s Marey, which claims to be trained exclusively on licensed data.